Distribution network adjustable resource regulation and control method based on adaptive density peak clustering and fractal PSO

Abstract

The invention provides a distribution network adjustable resource regulation and control method based on adaptive density peak clustering and fractal PSO, and relates to the technical field of power systems, and the method comprises the steps: obtaining the operation parameters of industrial adjustable resources in a distribution network, converting the operation parameters into a multi-dimensional feature matrix, determining a clustering center according to the local density and the relative distance of the data points in the multi-dimensional feature matrix to obtain a resource potential level; solving a resource demand matching model through a particle swarm algorithm based on a multi-scale disturbance quantity in combination with the resource potential level, and generating an intelligent contract; correcting an electrical distance between nodes in the power distribution network by using the resource potential level, and performing region division on the power distribution network based on the corrected electrical distance to obtain a region division result; and inputting the region division result, the matching result and the intelligent contract into a double-layer solving model to generate a regulation and control instruction. According to the method, the whole process integration of the distribution network adjustable resources from potential evaluation, optimization matching to partition regulation and strategy output is realized, and the method has high credibility and engineering applicability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power systems, in particular to a distribution network adjustable resource regulation method based on adaptive density peak clustering and fractal PSO. BACKGROUND

[0002] With the rapid access of distributed photovoltaic, wind power, energy storage and various types of industrial adjustable loads in the distribution network, the modern distribution network operation presents the characteristics of high distribution, multi-source heterogeneity and dynamic uncertainty. On the one hand, in the electrical layer, the node voltage, current and power fluctuation is significantly enhanced, and the traditional scheduling method relying on static load model cannot accurately depict the adjustable characteristics of different types of industrial loads, resulting in that the resource potential evaluation is not fine enough, and then affecting the rationality and economy of the regulation scheme. On the other hand, in the control layer, the scheduling system needs to issue tasks and coordinate responses among a large number of nodes, but the existing scheduling mechanism usually relies on centralized control, which has the problems of opaque task allocation process, difficult to trace resource response and low cross-regional coordination efficiency.

[0003] The existing adjustable resource evaluation and regulation method still has obvious deficiencies: the potential evaluation method is mostly based on fixed indicators or traditional clustering algorithms, which is difficult to adaptively select clustering centers and threshold values, resulting in unstable resource classification results, which weakens the accuracy of subsequent regulation; in the resource-demand matching link, the existing method mostly relies on heuristic optimization or centralized scheduling, lacks comprehensive consideration of resource dynamic characteristics and credible execution, and is difficult to guarantee global optimality and verifiability of execution; at the same time, in the partition and hierarchical regulation, the adjustable resource potential characteristics are not integrated into the regional modeling, resulting in large cross-regional regulation loss and insufficient utilization of key resources; in the regulation strategy output process, the existing method mostly fails to effectively handle load prediction errors and resource execution deviations, lacks robustness and dynamic adaptation ability. SUMMARY

[0004] The present application provides a distribution network adjustable resource regulation method based on adaptive density peak clustering and fractal PSO to solve the problems of inaccurate adjustable resource potential evaluation, low cross-regional regulation efficiency and insufficient robustness of regulation strategy in the operation of distribution network.

[0005] In the first aspect of the present application, a distribution network adjustable resource regulation method based on adaptive density peak clustering and fractal PSO is provided, comprising: obtaining the operation parameters of industrial adjustable resources in the distribution network, converting the operation parameters into a multi-dimensional feature matrix, determining the clustering center according to the local density and relative distance of the data points in the multi-dimensional feature matrix, and obtaining the resource potential level; By using a particle swarm algorithm based on multi-scale disturbance quantity, the resource demand matching model is solved in combination with the resource potential level, the matching result is obtained, and an intelligent contract is generated; The resource potential level is used to correct electrical distances between nodes in the power distribution network, and regional division is performed on the power distribution network based on the corrected electrical distances to obtain a regional division result. The regional division result, the matching result, and the smart contract are input into a double-layer solution model to generate a regulation and control instruction.

[0006] Optionally, in a possible implementation manner of the first aspect, the operation parameter includes maximum adjustable power, minimum adjustable power, power adjustment rate, allowed interruption duration, and past regulation and control success rate. The operation parameter is converted into a multi-dimensional feature matrix, including: The operation parameter is subjected to normalized mapping calculation to obtain a normalized feature vector. The normalized feature vectors of all industrial adjustable resources in the power distribution network are subjected to combination splicing processing to obtain the multi-dimensional feature matrix.

[0007] Optionally, in a possible implementation manner of the first aspect, the cluster center is determined according to local density and relative distance of a data point in the multi-dimensional feature matrix, including: The Euclidean distance between any two data points in the multi-dimensional feature matrix is determined, and sorting processing is performed on all the Euclidean distances. The distance located at a preset proportion position in the sorting result is selected, and a cut-off distance is determined based on the selected distance. The number of data points in the neighborhood of each data point is counted based on the cut-off distance to obtain the local density. The minimum distance from each data point to a data point having a local density greater than the current data point is determined to obtain the relative distance. Numerical fusion calculation is performed based on the local density and the relative distance to obtain a clustering decision value. The data point with the clustering decision value greater than a preset screening threshold is determined as the cluster center.

[0008] Optionally, in a possible implementation manner of the first aspect, the resource demand matching model is solved by the particle swarm algorithm based on the multi-scale perturbation quantity in combination with the resource potential level, including: A random sequence with a self-similar structure is generated by using a fractal function. The random sequence is processed based on an amplitude parameter to obtain the multi-scale perturbation quantity. In an iteration process of the particle swarm algorithm, the position data of a particle is updated by using the multi-scale perturbation quantity.

[0009] Optionally, in a possible implementation manner of the first aspect, the smart contract comprises a resource identity, a matching power, a regulation period, a cost settlement rule, and a default handling parameter. The method further comprises: During execution of the smart contract, an actual response power of the industrial adjustable resource is monitored, and an execution error is obtained by comparing the actual response power with the matching power. When the execution error exceeds a preset range, a reputation score of the industrial adjustable resource on the blockchain is adjusted, and the resource potential level is updated by using the adjusted reputation score.

[0010] Optionally, in a possible implementation manner of the first aspect, the electrical distance between nodes in the power distribution network is corrected by using the resource potential level, comprising: A node impedance matrix of the power distribution network is obtained, and an equivalent impedance module value between nodes is extracted as an initial electrical distance; A correction parameter is constructed, and a mapping relationship between the correction parameter and the resource potential level of the industrial adjustable resource to which the node belongs is established; The initial electrical distance is corrected based on the correction parameter, and a potential-aware electrical distance is obtained.

[0011] Optionally, in a possible implementation manner of the first aspect, the power distribution network is divided into regions based on the corrected electrical distance, and a region division result is obtained, comprising: A voltage sensitivity is obtained by determining a response value of a voltage amplitude variation of each node in the power distribution network with respect to an active power injection variation and a reactive power injection variation; A region division objective function is established, and the region division objective function takes minimizing an accumulated result of the potential-aware electrical distance between nodes in a region and maximizing an accumulated result of the voltage sensitivity of the nodes in the region as an optimization objective; The region division objective function is solved, and the region division result is obtained.

[0012] Optionally, in a possible implementation manner of the first aspect, the double-layer solving model comprises an upper-layer global scheduling model and a lower-layer region execution model; The upper-layer global scheduling model takes minimizing a total regulation cost of the power distribution network and a cross-region power exchange amount as an objective, and outputs a total amount of power regulation tasks of each region; The lower-layer region execution model takes minimizing a regulation power error and a response delay time in the region as an objective, and distributes the total amount of power regulation tasks to the industrial adjustable resource.

[0013] Optionally, in a possible implementation manner of the first aspect, the double-layer solving model comprises a contract constraint condition and an opportunity constraint condition. The contract constraint condition defines that the actual execution power upper limit of the industrial adjustable resource is determined by the physical adjustable upper limit and a contract credibility parameter; The opportunity constraint condition defines that, under the condition of considering the probability distribution of wind and light output prediction error and load fluctuation uncertainty, the probability of satisfying power balance of the power distribution network is greater than a preset confidence level.

[0014] Optionally, in a possible implementation manner of the first aspect, the data input into the double-layer solving model further includes a resource priority sequence, and a generation process of the resource priority sequence includes: obtaining a contract credibility, a response speed and a regulation cost of the industrial adjustable resource; solving a multi-objective optimization problem to obtain a set of importance parameters, the multi-objective optimization problem being aimed at minimizing priority distribution dispersion and minimizing power distribution network regulation cost; performing comprehensive evaluation and calculation on the contract credibility, the response speed and the regulation cost based on the set of importance parameters to obtain a priority score, and performing sorting processing on the industrial adjustable resource according to the priority score to obtain the resource priority sequence.

[0015] In a second aspect, the application provides an electronic device, including a memory, a processor and a computer program, the computer program is stored in the memory, and the processor runs the computer program to execute the method of the first aspect and various possible aspects related to the method.

[0016] The adaptive density peak clustering and fractal PSO distribution network adjustable resource regulation method provided by the application has the following beneficial effects: 1. In the potential evaluation, the application adaptively selects a clustering center and a truncation distance, realizes fine quantization and hierarchical division of potential of multiple industrial adjustable resources, and combines global multi-scale search and local convergence characteristics through a fractal PSO optimization mechanism, and deeply couples the intelligent contract to realize optimal matching and trusted execution of resource-demand pairs.

[0017] 2. The application combines electrical distance and sensitivity analysis with resource potential grade through the establishment of a potential-aware zoned hierarchical regulation framework, reduces cross-zone power coupling loss, and improves utilization efficiency of high-potential resources in the region.

[0018] 3. The application forms a closed-loop regulation link from potential evaluation, optimal matching to strategy output by integrating system economy, regional executability and uncertainty robustness into a unified framework through a double-layer optimization model based on contract constraints and opportunity constraints.

[0019] In conclusion, the application realizes adaptive evaluation, reliable matching, zoned and graded regulation and robust strategy output of the adjustable resources in the power distribution network under the operation environment, improves the reliability, flexibility and economy, and provides a systematic and deployable solution for the safe and efficient operation of the power distribution network under the background of large-scale distributed resource access, which has outstanding engineering application value and popularization significance. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 is a flowchart of the adjustable resource regulation method of the power distribution network provided by the adaptive density peak clustering and fractal PSO provided by the embodiments of the application; Figure 2 is a schematic diagram of the resource matching mechanism of the adjustable resource regulation method of the power distribution network provided by the adaptive density peak clustering and fractal PSO provided by the embodiments of the application; Figure 3 is a double-layer optimization model schematic diagram of the adjustable resource regulation method of the power distribution network provided by the adaptive density peak clustering and fractal PSO provided by the embodiments of the application; Figure 4 is a hardware structure schematic diagram of an electronic device provided by the embodiments of the application. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical scheme and advantages of the embodiments of the application clearer, the technical scheme in the embodiments of the application will be described clearly and completely in combination with the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0022] The technical scheme of the application will be described in detail in specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes can not be described in some embodiments.

[0023] Referring to Figure 1 is a flowchart of the adjustable resource regulation method of the power distribution network provided by the adaptive density peak clustering and fractal PSO provided by the embodiments of the application, Figure 1The execution subject of the method shown can be a software and / or hardware device. The execution subject of the present application can include, but is not limited to, at least one of the following: user equipment, network equipment, and the like. Among them, the user equipment can include, but is not limited to, a computer, a smart phone, a personal digital assistant (PDA), and the above-mentioned electronic devices, and the like. The network equipment can include, but is not limited to, a single network server, a server group composed of multiple network servers, or a cloud composed of a large number of computers or network servers based on cloud computing, wherein cloud computing is a kind of distributed computing, which is a super virtual computer composed of a group of loosely coupled computers. The present embodiment does not make any limitation. Steps 100 to 400 are included, and the details are as follows: Step 100: Obtain the operation parameters of the industrial adjustable resources in the power distribution network, convert the operation parameters into a multi-dimensional feature matrix, determine the cluster center according to the local density and relative distance of the data points in the multi-dimensional feature matrix, and obtain the resource potential level.

[0024] It should be noted that, in view of the problems that the operation characteristics of the massive industrial adjustable resources in the power distribution network are complex, the process constraints are various, and the regulation and control capacity is difficult to be quantified, the present application embodiment forms a multi-dimensional feature matrix capable of comprehensively representing the adjustable power range, dynamic response capacity and process constraint conditions of the resources through steps such as multi-source operation data acquisition, feature vectorization, normalization and abnormal correction, thereby providing basic data support for subsequent potential evaluation and intelligent matching.

[0025] In some embodiments, the specific implementation of step 100 includes steps 110 and 120: Step 110: Obtain the operation parameters of the industrial adjustable resources in the power distribution network, and convert the operation parameters into a multi-dimensional feature matrix.

[0026] It should be noted that the operation parameters include maximum adjustable power, minimum adjustable power, power regulation rate, allowed interruption duration, and past regulation and control success rate.

[0027] Specifically, for various types of industrial adjustable loads distributed in the power distribution network, a multi-source data set covering electrical parameters, process constraint parameters and historical behavior parameters is established. Specifically, it includes: maximum adjustable power , minimum adjustable power , rated power , power regulation rate , allowed interruption duration , process constraint coefficient , historical regulation and control success rate , and average response delay Key indicators. By joint collection of equipment operation monitoring system, energy consumption management system and historical scheduling log, a raw data set containing electrical characteristics, dynamic response capability and process constraint conditions is formed, thereby realizing multi-dimensional comprehensive characterization of resource operation state.

[0028] Specifically, the conversion of the operation parameters into the multi-dimensional feature matrix in step 110 includes steps 111 and 112: Step 111: performing normalization mapping calculation on the operation parameters to obtain a normalized feature vector.

[0029] The multi-source parameters of various industrial adjustable resources are uniformly modeled to construct a structured feature vector: ; Each dimension corresponds to adjustable power boundary, rated capacity, adjustment speed, allowable interruption, process constraint strength, historical regulation reliability and response delay, etc. The vectorization expression realizes the unified structured processing of heterogeneous data, so that the clustering analysis in the subsequent application can be calculated and compared in a unified feature space, effectively solving the problem of incommensurability caused by large differences in industrial load and inconsistent attributes.

[0030] Further, in view of the dimensional difference and inconsistent numerical range of different indicators, a min-max normalization method is introduced to convert the feature vector into dimensionless data after normalization: ; Wherein, represents the resource The original value in the first dimension, represents all values of this dimension data, represents the normalized result. For missing values in the collection process, historical mean interpolation or adjacent interpolation method is used for correction; for abnormal values, a combination of three standard deviation test and box plot test is used to identify and eliminate. After the above processing, a clean, unified and directly usable feature vector set is obtained.

[0031] Step 112: performing combined splicing processing on the normalized feature vectors of all industrial adjustable resources in the power distribution network to obtain a multi-dimensional feature matrix.

[0032] The normalized feature vectors of all resources are combined into a multi-dimensional feature matrix : ; Wherein, is the number of resources, is the normalized feature vector, For feature dimension.

[0033] It should be noted that the multi-dimensional feature matrix not only uniformly represents the adjustable capacity of the resource and the process constraint, but also integrates the historical execution performance and the response characteristics, realizes the fusion modeling of static parameters and dynamic behaviors. The finally output feature matrix provides high-quality input data for potential assessment, so that the assessment result can accurately reflect the real regulation potential of the resource, and lay a solid foundation for the resource matching driven by the intelligent contract.

[0034] Step 120: determining the clustering center according to the local density and relative distance of the data points in the multi-dimensional feature matrix.

[0035] It should be noted that, for the obvious differences in the adjustable power range, response speed and process constraint of the massive industrial adjustable resources in the power distribution network, the traditional evaluation method is difficult to accurately identify the resource potential level, therefore, the application comprehensively considers the interruptibility, transferability, economy and reliability of the resource in the multi-dimensional feature space, and realizes the automatic hierarchical division of the adjustable resource potential by adaptively determining the clustering key parameters and the center point.

[0036] Specifically, step 120 includes steps 121 to 126: Step 121: determining the Euclidean distance between any two data points in the multi-dimensional feature matrix, and performing sorting processing on all Euclidean distances.

[0037] Step 122: selecting the distance located at the preset proportion position in the sorting result, and determining the cut-off distance based on the selected distance.

[0038] Step 123: based on the cut-off distance, the number of data points in the neighborhood of each data point is counted to obtain the local density.

[0039] It should be noted that, for each resource point in the multi-dimensional feature matrix , that is, the data point, the local density is defined as: ; Wherein, denotes the Euclidean distance between the resource and the resource , and the cut-off distance is . The local density reflects the concentration degree of the neighboring resources around the resource point, and the higher it is, the more representative it is in the adjustable power, response speed and process constraint, and thus it is more likely to become a high-level candidate in the potential classification.

[0040] Further, all Euclidean distances, that is, all sample spacings, are sorted to avoid the uncertainty caused by manually setting the cut-off distance. The coverage proportion threshold (i.e., preset proportional position), and before The mean of the minimum distances is used as the cutoff distance: ; in, , Indicates the first Small sample intervals. This application can automatically adjust the calculation scale of local density according to resource distribution, ensuring that clustering can still accurately reflect the actual potential level of adjustable resources in scenarios where resource aggregation and sparsity coexist.

[0041] Step 124: Determine the minimum distance from each data point to a data point with a local density greater than the current data point, and obtain the relative distance.

[0042] Step 125: Perform numerical fusion calculation based on local density and relative distance to obtain clustering decision values.

[0043] It should be noted that, based on the local density, the minimum distance from each resource point to a point with higher density is calculated: ; in, Representing resources The relative distance. Taking into account both local density and relative distance, a clustering index is defined as: ; when When the value is large, the resource is not only highly representative in terms of potential, but also maintains sufficient differences from other high-potential resources. Therefore, it is adaptively identified as a cluster center, which can ensure that the evaluation results can effectively distinguish core resource groups of different potential levels.

[0044] Step 126: Determine the data points whose clustering decision value is greater than the preset screening threshold as cluster centers.

[0045] A preset screening threshold is a numerical limit used to determine whether a clustering decision value (i.e., the product of local density and relative distance) is significant. Data points exceeding this preset screening threshold are identified as cluster centers that can represent specific potential clusters because they simultaneously possess high local density and high relative distance characteristics. The preset screening threshold is usually adaptively determined based on the ranking distribution characteristics of the clustering decision values ​​of all data points, aiming to automatically identify extreme resources with decision values ​​significantly higher than other points as cluster centers.

[0046] Each resource is sequentially assigned to its nearest high-density neighbor, forming multiple potential clusters. For each cluster, resources are classified into three potential levels: high, medium, and low, based on the potential level of its cluster center and the overall performance within the cluster. ; Among them, the resources in the high-potential cluster often have greater adjustable power, faster response speed and higher reliability, the medium-potential cluster is a compromise between regulation ability and constraint condition, and the low-potential cluster is suitable for backup or supplementary regulation scenarios. Therefore, the output potential level can not only intuitively reflect the value difference of resources in regulation, but also provide hierarchical basis for smart contract matching and priority regulation sorting; the adaptive, automatic hierarchical evaluation of the potential of adjustable resources is realized, the deviation caused by manual intervention is avoided, the stability and robustness of the evaluation results under different scale and characteristic distribution conditions are ensured, and thus the scientificity and executability of the intelligent regulation of the distribution network adjustable resources are improved.

[0047] Step 200: Solving the resource demand matching model by combining the resource potential level based on the particle swarm algorithm based on multi-scale disturbance quantity, obtaining the matching result and generating the smart contract.

[0048] It should be noted that, in view of the multi-dimensional matching constraints between the adjustable resources and the regulation and control demand of the distribution network in terms of capacity, response time and economy, the traditional matching method is prone to local optimum and lacks reliable guarantee in execution, and a resource matching mechanism based on fractal PSO and smart contract integration is proposed, as shown in Figure 2 The resource matching mechanism introduces the fractal genetic principle in the search process, realizes the balance of global-local multi-scale search, and solidifies the optimal matching result in the form of a smart contract on the blockchain, thereby realizing the efficiency and reliability of resource matching.

[0049] An optimization model of resource-demand matching is established, and the resource potential level, adjustable power , response delay , process constraint and economic cost are taken as decision variables. The objective function is defined as the maximization of the comprehensive benefit after matching or the minimization of the matching cost: ; Wherein represents whether the resource is selected, is a processing factor. The objective function can take into account economy, response speed and process constraint at the same time.

[0050] Specifically, the resource demand matching model is solved by combining the resource potential level based on the particle swarm algorithm based on multi-scale disturbance quantity, including steps A1 to A3: Step A1: generating a random sequence with self-similar structure by using a fractal function.

[0051] Step A2: performing processing on the random sequence based on the amplitude parameter to obtain multi-scale disturbance quantity.

[0052] Step A3: In the iteration process of the particle swarm algorithm, the position data of the particles is updated by using the multi-scale perturbation quantity.

[0053] It should be noted that the fractal genetic principle is introduced into the speed and position updating mechanism of the traditional PSO algorithm: ; ; wherein, represents the position of the particle at iteration (i.e., the matching scheme code), represents its speed and are the individual historical optimal and global optimal positions, respectively; is the inertia weight; is the learning factor; , represents r 1 and r 2 are random numbers subject to uniform distribution in [0,1], represents uniform distribution in the interval [0,1], represents the position of the particle at iteration . The function represents a fractal perturbation operator, which generates multi-scale perturbations based on self-similar structures to introduce search space with a magnitude factor .

[0054] Through the fractal perturbation mechanism, the particle swarm maintains a dynamic balance between global exploration and local development, reducing the risk of falling into a local optimum.

[0055] It should be noted that the smart contract includes resource identity, matching power, regulation period, cost settlement rule, and default handling parameters.

[0056] In some embodiments, the present application further comprises steps B1 and B2: Step B1: During the execution of the smart contract, the actual response power of the industrial adjustable resource and the matching power are monitored to obtain an execution error.

[0057] Step B2: In response to the execution error exceeding the preset range, the reputation score of the industrial adjustable resource on the blockchain is adjusted, and the resource potential level is updated using the adjusted reputation score.

[0058] The preset range refers to the maximum deviation tolerance interval allowed for the actual response power of the industrial adjustable resource relative to the target matching power set in the smart contract terms. The preset range is usually determined when the contract template is generated in the resource matching stage, based on the requirements of the power distribution network for regulation accuracy and the process constraint characteristics of the industrial resource.

[0059] It should be noted that the optimal matching result obtained by the fractal PSO is converted into a smart contract template, which contains resource number, matching power , regulation period , cost settlement rule and default handling parameters. The contract is deployed on the chain, and the consensus mechanism is used to ensure its tamper resistance, and the regulation instruction is automatically executed when the trigger condition is met. The smart contract not only eliminates the delay and uncertainty of manual execution, but also provides transparent and traceable execution guarantee.

[0060] During the contract execution process, the response of the resource is automatically recorded, including the success rate of execution, the actual response time and the power deviation. If the execution result does not meet the expectation, the resource reputation level will be lowered and fed back to the potential evaluation module, so as to reduce its priority in the next round of clustering and matching, thereby realizing the dynamic optimization process of evaluation-matching-execution-correction to improve the adaptive ability and long-term regulation performance.

[0061] Step 300: Correct the electrical distance between nodes in the power distribution network by using the resource potential level, and perform regional division of the power distribution network based on the corrected electrical distance to obtain a regional division result.

[0062] It should be noted that the present application aims at the problem of uneven spatial distribution of adjustable resources in the power distribution network and significant differences in electrical coupling between regions. The potential evaluation result and the resource-demand matching scheme are used as inputs, the potential constraint and the contract execution mechanism are introduced in the regional division, which not only guarantees the electrical coupling density, but also strengthens the utilization rate of high-potential resources within the region, and realizes multi-level coordinated control from global scheduling to regional execution through hierarchical regulation model, thereby improving the execution efficiency and robustness of the regulation scheme.

[0063] In some embodiments, the specific implementation of step 300 includes step 310 and step 320: Step 310: Correct the electrical distance between nodes in the power distribution network by using the resource potential level.

[0064] Specifically, step 310 includes steps 311 to 313: Step 311: Obtain the node impedance matrix of the power distribution network, and extract the equivalent impedance modulus value between nodes as the initial electrical distance.

[0065] Step 312: Construct the correction parameter, and establish the mapping relationship between the correction parameter and the resource potential level of the industrial adjustable resource to which the node belongs.

[0066] Step 313: Perform correction calculation on the initial electrical distance based on the correction parameter to obtain the potential-aware electrical distance.

[0067] It should be noted that the power distribution network node set is , and the branch impedance matrix is . The electrical distance between node and node is defined as: ; wherein represents the equivalent impedance of node and in the impedance matrix, represents the self-impedance of node . In order to reflect the difference in resource potential, a resource potential level correction factor is introduced, and the potential-aware electrical distance between node and node is constructed as: ; wherein represents the average potential level normalized value of the resource on node (obtained by clustering in the foregoing step), is a weight coefficient; the correction calculation of the present application can preferentially aggregate high-potential resources in the partition process, and ensure that the key area has stronger rapid response capability.

[0068] Step 320: performing regional division on the power distribution network based on the corrected electrical distance, to obtain a regional division result.

[0069] Specifically, step 320 includes steps 321 to 323: Step 321: determining the response value of the voltage amplitude variation of each node in the power distribution network with respect to the active power injection variation and the reactive power injection variation, to obtain the voltage sensitivity.

[0070] On the basis of the potential-aware electrical distance, a sensitivity index of the node voltage to active and reactive power injection is further introduced, which is defined as: ; wherein represents the voltage amplitude of node , are the active and reactive power injections of node respectively. The greater the sensitivity index , the more sensitive the state of node to the regulation behavior of node , and node should be preferentially allocated to the same region to reduce energy loss and response delay caused by cross-region regulation.

[0071] Step 322: Establish a regional division objective function, which aims to minimize the cumulative potential-aware electrical distance between nodes within a region and maximize the cumulative voltage sensitivity of nodes within a region.

[0072] Step 323: Perform solving on the regional division objective function to obtain the regional division result.

[0073] It should be noted that the potential-aware electrical distance and sensitivity index are integrated to construct the regional division optimization objective, and the regional division of the distribution network is performed through a weighted clustering method: ; Wherein, , represents whether the node and the node are divided into different regions, represents the th sub-region, is a weight parameter.

[0074] Step 400: Input the regional division result, matching result and smart contract into the double-layer solving model to generate the regulation and control instruction.

[0075] In view of the problems of lack of global consistency in the output process of traditional regulation and control strategy, difficulty in balancing safety and economy, and lack of closed-loop feedback in the execution link, an optimal regulation and control strategy output method based on contract constraint is proposed. With the potential assessment, fractal PSO matching result, zoned hierarchical regulation modeling and dynamic priority sorting in the foregoing steps as inputs, a double-layer optimization model is constructed, and a smart contract is combined to realize credible execution and result backtracking, so as to ensure the optimality and robustness of the regulation and control strategy under complex environment, as shown in Figure 3 .

[0076] On the basis of regional division, a zoned hierarchical regulation framework is established.

[0077] (1) Global scheduling layer: based on the fractal PSO matching result, the global optimal resource-demand pair is determined, and the regulation and control task is issued to each region.

[0078] (2) Regional control layer: after receiving the global task, each regional controller combines the potential distribution of the region to further decompose the global task, and issues the task in the region in the form of a smart contract to ensure credible execution.

[0079] (3) Terminal resource layer: according to the execution content of the regional contract, the specific regulation and control is completed, and the execution situation is automatically recorded on the chain and fed back to the upper layer.

[0080] The contract execution mechanism is extended to the regional level to complete the integrated hierarchical regulation of global-region-terminal.

[0081] Within each region, resources with different potential levels are called in a hierarchical manner: high-potential resources are preferentially assigned to fast and critical tasks; medium-potential resources are responsible for balancing and auxiliary control; and low-potential resources are used as redundancy backup, mainly for coping with uncertain deviations.

[0082] Further, the double-layer solving model includes an upper-layer global scheduling model and a lower-layer regional execution model.

[0083] The upper-layer global scheduling model aims to minimize the total control cost of the power distribution network and the cross-regional power exchange amount, and outputs the total amount of power regulation tasks for each region.

[0084] The lower-layer regional execution model aims to minimize the control power error and response delay time in the region, and allocates the total amount of power regulation tasks to the industrial adjustable resources.

[0085] Further, the double-layer solving model includes contract constraints and opportunity constraints.

[0086] The contract constraints limit the actual execution power upper limit of the industrial adjustable resources, which is determined by the physical adjustable upper limit and the contract credit parameter.

[0087] The opportunity constraints limit the probability of meeting the power balance of the power distribution network under the consideration of the probability distribution of the wind and light output prediction error and the load fluctuation uncertainty to be greater than the pre-set confidence level. The pre-set confidence level refers to the minimum probability threshold that must be met to satisfy the power balance constraint in the double-layer optimization model to cope with the uncertainty of wind and light output prediction error and industrial load fluctuation. For example, if the confidence level is set to 95% (corresponding to the risk tolerance in the technical disclosure ), it means that considering the randomness of renewable energy output and load fluctuation, the generated control strategy must ensure that the system has at least a 95% probability of maintaining power supply and demand balance, without over-limit or load shedding.

[0088] It should be noted that the optimal control problem is modeled as a double-layer optimization problem: (1) Upper-layer optimization (global scheduling layer): the goal is to minimize the total control cost of the system, and the constraint conditions include system power balance, line flow limit, and frequency and voltage stability requirements.

[0089] (2) Lower-layer optimization (regional execution layer): in each region, the resource calling order is determined according to the priority sequence , and the goal is to minimize the control deviation and execution delay.

[0090] Contract constraints are introduced in the double-layer optimization: ; wherein, For resources Execution power, Set its maximum callable limit. This is a contractual credibility factor. If a resource has a history of default, its... This reduces the amount of tasks that can be allocated, thereby ensuring that the optimized solution not only considers technical constraints but also incorporates the credibility of contract execution.

[0091] To address the forecasting errors in wind and solar power output and the uncertainties in industrial load, opportunity constraints are established: ; in, As backup capacity, For the required power, This represents risk tolerance. Opportunity constraints ensure that load demands can be met with a high probability, even under extreme uncertainty.

[0092] In some embodiments, the data input to the two-layer solution model further includes a resource priority sequence, and the process of generating the resource priority sequence includes steps C1 to C3: Step C1: Obtain the contractual credibility, response speed, and control costs of industrial adjustable resources.

[0093] Step C2: Solve the multi-objective optimization problem to obtain the set of importance parameters. The objectives of the multi-objective optimization problem are to minimize the priority distribution dispersion and minimize the distribution network regulation cost.

[0094] Step C3: Perform a comprehensive evaluation calculation on contract credibility, response speed and control cost based on the set of importance parameters to obtain priority scores. Sort the industrial adjustable resources according to the priority scores to obtain the resource priority sequence.

[0095] It should be noted that, assuming the first The priority score of each resource is Considering potential level Response speed Economic costs ,reliability and contract credibility Define the priority function: ; in, This is an adaptive weighting mechanism. Unlike traditional methods, this mechanism is the first to... (Obtained from contract execution feedback in the aforementioned steps) By incorporating ranking indicators, the priority not only depends on static attributes but also reflects historical contract performance, thereby enhancing the credibility of regulation.

[0096] Weight vector By multi-objective optimization in real time, the objective function is: ; Wherein, represents the dispersion of priority distribution, represents the system regulation cost (summed by the regional cost function in the foregoing step), is the adjustment parameter. The application automatically adjusts the weight under different operating scenarios, avoiding the imbalance of the ranking caused by artificial static setting.

[0097] In each regulation cycle, the resource priority sequence is updated according to the following dynamic rules: ; Wherein, is the execution deviation of the resource, is the response delay. If the deviation of a resource exceeds the limit or violates the contract during execution, its priority will obviously decrease in the next round of ranking. The dynamic rules ensure that the priority ranking can reflect the reliability and stability of the resource in real time.

[0098] In some embodiments, the data input into the double-layer solving model also includes a resource priority sequence, and the generation process of the resource priority sequence further includes step C4: considering the load prediction error and uncertainty disturbance, and performing correction on the resource priority sequence by using a robust ranking method based on chance constraints.

[0099] Specifically, a priority threshold and a tolerable violation probability are set to ensure the stability of the ranking of the resource under uncertainty conditions, and a final robust priority sequence is obtained.

[0100] The final regulation strategy is published on the chain in the form of a smart contract, and the regional controller and the terminal resource execute according to the contract terms, and the execution data and deviation results are automatically recorded on the chain. The non-tamperability of the blockchain ensures the transparency and traceability of the strategy execution. The execution results will be fed back to step 100 and the generation process of the resource priority sequence, respectively, for dynamic updating of the potential level and evolution of the priority sequence, thereby forming a closed-loop optimization mechanism of regulation-execution-feedback-regulation.

[0101] On the basis of the foregoing steps, the application further includes the following embodiments: Referring to Figure 4 is a hardware structure schematic diagram of an electronic device provided by an embodiment of the application, the electronic device 40 comprising: a processor 41, a memory 42, and a computer program; wherein, The memory 42 is configured to store the computer program, and the memory can also be a flash memory. The computer program is, for example, an application program, a functional module, etc. that implements the foregoing method.

[0102] The processor 41 is configured to execute the computer program stored in the memory to implement the steps performed by the device in the above method. Details can be referred to the description of the above method embodiments.

[0103] Optionally, the memory 42 can be independent or integrated with the processor 41.

[0104] When the memory 42 is independent of the processor 41, the device can further include: A bus 43 is configured to connect the memory 42 and the processor 41.

[0105] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, but not limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for adjustable resource regulation in distribution networks based on adaptive density peak clustering and fractal PSO, characterized in that, include: The operating parameters of industrial adjustable resources in the distribution network are obtained, the operating parameters are converted into a multi-dimensional feature matrix, and the cluster center is determined according to the local density and relative distance of data points in the multi-dimensional feature matrix to obtain the resource potential level. By using a particle swarm optimization algorithm based on multi-scale perturbation, combined with the resource potential level, a resource demand matching model is solved to obtain the matching results and generate a smart contract. The electrical distances between nodes in the distribution network are corrected using the resource potential level, and the distribution network is divided into regions based on the corrected electrical distances to obtain the region division results. The region division results, the matching results, and the smart contract are input into a two-layer solution model to generate control instructions.

2. The method according to claim 1, characterized in that, The operating parameters include maximum adjustable power, minimum adjustable power, power adjustment rate, allowable interruption duration, and past adjustment success rate; The step of converting the operating parameters into a multidimensional feature matrix includes: Perform normalized mapping calculations on the operating parameters to obtain normalized feature vectors; The normalized feature vectors of all industrial adjustable resources in the distribution network are combined and spliced ​​to obtain the multidimensional feature matrix.

3. The method according to claim 2, characterized in that, The step of determining cluster centers based on the local density and relative distance of data points in the multidimensional feature matrix includes: Determine the Euclidean distance between any two data points in the multidimensional feature matrix, and sort all the Euclidean distances. Select the distance in the sorting results that is located at a preset proportion position, and determine the cutoff distance based on the selected distance; The local density is obtained by counting the number of data points in the neighborhood of each data point based on the cutoff distance; The minimum distance from each data point to a data point with a local density greater than that of the current data point is determined to obtain the relative distance; Numerical fusion calculations are performed based on the local density and the relative distance to obtain clustering decision values; Data points whose clustering decision values ​​are greater than a preset screening threshold are identified as cluster centers.

4. The method according to claim 1, characterized in that, The process of solving the resource demand matching model using the particle swarm optimization algorithm based on multi-scale perturbations, combined with the resource potential level, includes: Generating random sequences with self-similar structures using fractal functions; The random sequence is processed based on the amplitude parameter to obtain the multi-scale perturbation amount; During the iterative process of the particle swarm optimization algorithm, the position data of the particles are updated using the multi-scale perturbation.

5. The method according to claim 4, characterized in that, The smart contract includes resource identity identifiers, matching power, control period, cost settlement rules, and default handling parameters. The method further includes: During the execution of the smart contract, the actual response power of the industrial adjustable resources and the matching power are monitored to obtain the execution error; When the execution error exceeds a preset range, the reputation score of the industrial adjustable resource on the blockchain is adjusted, and the resource potential level is updated using the adjusted reputation score.

6. The method according to claim 1, characterized in that, The method of using the resource potential level to correct the electrical distance between nodes in the distribution network includes: Obtain the node impedance matrix of the distribution network and extract the equivalent impedance magnitude between nodes as the initial electrical distance; Construct correction parameters and establish a mapping relationship between the correction parameters and the resource potential level of the industrial adjustable resources to which the node belongs; Based on the correction parameters, a correction calculation is performed on the initial electrical distance to obtain the potential sensing electrical distance.

7. The method according to claim 6, characterized in that, The process of dividing the distribution network into zones based on the corrected electrical distance to obtain zone division results includes: The voltage sensitivity is obtained by determining the response values ​​of the voltage amplitude change at each node in the distribution network relative to the changes in active power injection and reactive power injection. A region partitioning objective function is established, which aims to minimize the cumulative result of the potential sensing electrical distance between nodes within the region and maximize the cumulative result of the voltage sensitivity of nodes within the region. The objective function for region partitioning is solved to obtain the region partitioning result.

8. The method according to claim 1, characterized in that, The two-layer solution model includes an upper-layer global scheduling model and a lower-layer regional execution model; The upper-level global scheduling model aims to minimize the total control cost of the distribution network and the amount of cross-regional power exchange, and outputs the total amount of power regulation tasks for each region. The lower-level regional execution model aims to minimize the regulation power error and response delay time within the region, and allocates the total amount of power regulation tasks to industrial adjustable resources.

9. The method according to claim 8, characterized in that, The two-layer solution model includes contractual constraints and opportunity constraints; The contractual constraints limit the actual execution power of industrial adjustable resources to an upper limit determined by the physical adjustable upper limit and the contractual reputation parameter. The opportunity constraint condition is limited to the probability distribution of the distribution network power balance being greater than the preset confidence level, taking into account the probability distribution of wind and solar power output prediction errors and load fluctuation uncertainties.

10. The method according to any one of claims 1, 8, or 9, characterized in that, The data input to the two-layer solution model also includes a resource priority sequence, the generation process of which includes: Contractual credibility, response speed, and control costs for acquiring industrially adjustable resources; Solving the multi-objective optimization problem yields a set of importance parameters. The objectives of the multi-objective optimization problem are to minimize the priority distribution dispersion and to minimize the distribution network regulation cost. Based on the set of importance parameters, a comprehensive evaluation calculation is performed on the contract credibility score, the response speed, and the regulation cost to obtain a priority score. The industrial adjustable resources are then sorted according to the priority score to obtain the resource priority sequence.

Images